Some questions about video.json in inference_single_pair_visuals.py

[RCpengnan]

1. -v1, -v2 : JSON path of the pose sequences of video 1 and video 2.
   Whether video1.json and video2 .json have different total frames? Whether they need to be processed into the same number of frames in advance before the experiment？
2. Whether the format of video1.json should be the same as that of S004C002P020R002A007.json.(S004C002P020R002A007.json from NTU RGB+D motion similarity annotations ). if so ,the b_score is body part score? Could you tell me how to get the b_score?
   .
   Thank you!